Resultados

My study focused on 4 state parks located in Northwestern Arkansas: Mount Magazine State Park, Mount Nebo State Park, Petit Jean State Park, and Pinnacle

Mountain State Park. Mount Magazine, Mount Nebo, and Petit Jean State Parks are located in the Arkansas River Valley ecoregion and Pinnacle Mountain State Park is located in the Ouachita Mountain ecoregion (USEPA 2016).

Mount Magazine State Park is located in Logan County, south of Paris, Arkansas (15 S 442199, 38952229) and encompasses 904ha surrounded by the Ozark National Forest. The park is positioned on top of the Magazine mountain (839m), a flat-topped plateau rimmed by sandstone bluffs which supports a diverse collection of wildlife and vegetation species adapted to the mountain ecosystem such as Ozark chinquapin (Castanea ozarkensis) and maple-leaf oak (Quercus acerifolia). The park area has a Linker fine sandy loam dominant soil type with small portions of gravelly clay loam surrounding the mountain (USDA and NRCS 2015). Mount Magazine had a mean minimum temperature of 18.3 ± 5.80^°C (mean ± 1 SD for all statistics hereafter) and a mean maximum temperature of 27.7 ± 5.80^°C during the study (Weather Underground 2015). Mean precipitation for the park during the study was 7.27 ± 6.54mm (Weather Underground 2015). The park included 8 trails spanning 22.5km in length, 13 cabins, and 18 campsites.

Mount Nebo State Park is located in Yell County, west of Dardanelle, Arkansas (15 S 476945, 3897552) and encompasses 1,246ha of habitat. The park is centered on top of Mount Nebo, which measures 411m in elevation. The habitat is mostly comprised of thick oak (Quercus spp.) and hickory (Carya spp.) dominated forests, characteristic of the Ozark Plateau region, with mixes of sweetgum (Liquidambar styraciflua) and red maple (Acer rubra) stands throughout the park. The park is dominantly a Nella-Enders stone fine sandy loam soil type with 8 – 20% slopes, with much of the mountain covered in

Linker-Mountainburg complex soil type (USDA and NRCS 2015). Mount Nebo had a mean minimum temperature of 21.4 ± 2.62^°C and a mean maximum temperature of 32.1

± 3.23^°C during the study. Mean precipitation for the park during the study was 8.33 ± 5.08mm. The park included 6 trails with a combined length of 22.5km, 15 cabins, and 44 campsites.

Petit Jean State Park is located in Conway County, west of Oppelo, Arkansas (15 S 505957, 3886563). Petit Jean mountain (368m) lies between the Ozark and Ouachita mountain ranges in the Arkansas River Valley and serves as the midpoint for the 1,416ha park. The habitat is comprised mostly of forests dominated by a mix of oak, hickory, and pine (Pinus spp.) stands within a series of ponds, streams, and glades, also characteristic of the Ozark mountain ecoregion (USEPA 2016). The park has a dominantly

Mountainburg stony fine sandy loam soil type with 12 – 40% slopes, with large patches of Linker fine sandy loam and Enders gravelly fine sandy loam surrounding the park (USDA and NRCS 2015). Petit Jean had a mean minimum temperature of 20.7 ± 2.16^°C and a mean maximum temperature of 32.1 ± 3.50^°C during the study. Mean precipitation for the park during the study was 1.87 ± 3.06mm. Eight trails traversing 37.0km, 33 cabins, and 125 campsites were available in the park.

Pinnacle Mountain State Park is located in Pulaski County, Northwest of Little Rock, Arkansas (15 S 547062, 3855665) and encompasses 809ha surrounding Pinnacle Mountain (308m). The park is composed of a mosaic of habitats including boulder fields, bald cypress (Taxodium distichum) swamps, bottomland hardwood forests, and upland forests composed of mixes of oak, hickory, and pine stands. The park supports a diverse variety of floral and faunal species as a result from the heterogeneity in habitat types

including the park’s Arboretum that contains woody vegetation from across the state and from the Big and Little Maumelle rivers that run through the park. The park is primarily of Carnassial-Mountainburg association undulating soil type, surrounded by patches of Perry clay soil type (USDA and NRCS 2015). Pinnacle Mountain had a mean minimum temperature of 23.2 ± 0.60^°C and a mean maximum temperature of 34.7 ± 2.42^°C during the study. Mean precipitation for the park during the study was 0.49 ± 0.84mm. Pinnacle Mountain State Park included 10 trails stretching 30.0km; no camping sites or cabins existed at the time of this study.

I focused data collection on specific trails chosen within each state park based on 3 criteria: total trail length (m), diversity of habitat types that the trail traversed, and the area of the park that each trail encompassed (km). I chose trails ≤16km in length and split trails measuring 8 – 16km into 2 equal portions to accommodate temporal limitations of accessing trails during each visit. I then used ArcGIS (Environmental Systems Research Institute, Inc., Redlands, CA) to overlay the trails meeting these distance criteria onto park maps and chose trails that passed through the greatest diversity of habitat types (USEPA 2016) and that represented the greatest area of park use. Applying these criteria to the trails in each park resulted in 6 trails each at Mount Magazine State Park, Mount Nebo State Park, and Petit Jean State Park and 8 trails at Pinnacle Mountain State Park.

I then created sampling points on the selected trail systems of each park because most human-wildlife interactions should occur on trails that promote recreational activities such as hiking and wildlife observation (Boyle and Samson 1985, Marion and Leung 2001, Cordell et al. 2008). For each trail, I created an initial sampling point randomly within the first 250m of the trailhead by generating a random distance (m) via

the sample program in Program R (sample=x, size, replace=FALSE, prob=NULL; R Version 3.1.2., http://www.r-project.org/). Subsequent sampling points were then systematically located every 250m (Ralph et al. 1995, Torn et al. 2009). This

methodology resulted in 227 sampling points for all 4 parks: Mount Magazine (n = 60);

Mount Nebo (n = 56); Petit Jean (n = 59); Pinnacle Mountain (n = 52).

Taxa Data Collection

I collected data in 1-week sessions from 18 May 2015 – 7 August 2015. During this time, 2 technicians and I sampled each park for 3 weeks. The reoccurring 1-week sessions followed this order: Mount Magazine State Park, Mount Nebo State Park, Petit Jean State Park, and Pinnacle Mountain State Park.

Avian point counts.— I conducted fixed distance point counts (see Ralph et al.

1995). Each point was sampled independently 3 times/week (Monday – Friday), once each by 3 observers (Petit et al. 1995). This methodology resulted in 9 visits for each of the 227 points (i.e., 3 times/week at each point during 3 independent weeks), with 45 minutes of total observation time collected/point. By utilizing 3 observers throughout the week rather than 1, as is common in many avian surveys, I was able to diminish repeated observer bias and increase the detection probability at each point (Ralph et al. 1995, MacKenzie and Royle 2005). No point was visited at the same time throughout the week by any of the 3 observers. Travel time between point counts was <10 minutes which reduced potential bias by avoiding repeated counts of avian individuals at adjacent points (Ralph et al. 1993). Counts began ≤15min of sunrise until 5 hours after sunrise (~0500 – 1000hr). Point counts lasted 5-min each and observers recorded both visual and auditory detections for birds within a 50m-radius. I identified birds to species level and used

4-letter alpha codes to record visual and auditory bird detections at each sampling point (Pyle and Desante 2003). Point counts were conducted only during suitable weather conditions for avian activity, which I defined as: mornings with no rain or fog (although temperate, light drizzle can be tolerated by most species; Cyr et al. 1995, Martin et al.

1997), wind speeds <13km/hr (Freedmark and Rogers 1995, Petit et al. 1995), and temperatures ranging 18 – 23^oC (Buskirk and McDonald 1995, Martin et al. 1997). If conditions did not meet these criteria or did not improve by 1000hr, counts were halted for that observation day and were continued during the next day or at the end of the sampling week to maintain independent samples (Ralph et al. 1993, Martin et al. 1997).

Mesocarnivore camera traps.— I used 8 Spypoint C4 camera traps (Spypoint, Swanton, Vermont) to collect community data for mesocarnivores in each park. Each camera was powered by 6 AA batteries and captured still photos via a motion-sensitive infrared trigger without the use of attractants (Shannon et al. 2014). I used a cable lock to mount cameras at a height of 30 – 50cm on trees parallel to the designated sampling point on the trail (Yasuda 2004, Gompper et al. 2006, TEAM 2011). Each camera was focused on the trail edge facing away from the trailhead to ensure captures of all individuals passing by (Harmsen et al. 2010). I adjusted the conical range of the camera sensors to a 5-m range resulting in a field of view of approximately 3.5-m to ensure photo clarity (TEAM 2011). I set cameras to a 3-min time delay between captures with 1 photo taken per trigger event to maximize photo amounts while balancing continuous sampling. This also avoided overfilling the 4GB SD card in each camera by waste footage caused by multiple records of a single animal or group of animals repeatedly triggering the trap (Yasuda 2004, Rowcliffe et al. 2008).

One camera was deployed at a randomly chosen sampling point on the Sunday of each sampling week between 1200 – 1500hrs and began recording photos by 1500hr that day. I programmed cameras to record 24-hours a day and to include the time and date when triggered until they were collected on Friday of the same week between 1200 – 1500hrs. After each sampling week, I identified captured mesocarnivore photos to the species level. I observed the time stamps of each photo capture to assure independent samples and to prevent the recording of multiple captures of the same animal during a single instance.

Woody vegetation subplots.— I used a nested subplot method to collect community data for understory and overstory woody vegetation similar to James and Shugart (1970) and the BBIRD monitoring program (Ralph et al. 1993, Martin et al.

1997). Vegetation data were collected at each sampling point once during the study. I centered circular subplots with 5-m and 11.3-m radii on adjusted sampling points established 16.3m away from each trail sampling point. This adjustment created a 5-m buffer between the edges of the trail and each vegetation plot to avoid immediate edge effects (Brown et al. 2009). I established these adjusted points randomly on either the left or right side of the trail; adjusted sampling points that were too dangerous to sample (e.g., close to the mountain edge or in a ravine) were either switched sides or conducted

without the 5-m buffer offset from the trail. Of the 227 sampling points, this methodology resulted in 97 sampling points adjusted left, 106 adjusted right, and 24 sampled on the trails.

I quantified all understory woody vegetation (saplings measuring ≤1.4m tall) within the 5-m plot and all overstory vegetation (trees measuring >1.4m tall) in the

11.3-m plot (Geldenhuys 1997, Rodewald and Brittingha11.3-m 2004, Brown et al. 2009). I also recorded canopy cover, slope, and aspect within the 5-m subplot for further

characterization of the vegetation community (Martin et al. 1997). I used a spherical densitometer to measure canopy cover by averaging cover recordings at 5m from center in all 4 cardinal directions at each sampling point. Percent slope was measured across the 5-m subplot from the top of the plot to bottom (facing downhill) with a clinometer and I recorded aspect in degrees with a compass at the top of the 5-m subplot facing downhill to record the azimuth of the slope.

Data Analyses

I calculated species composition, species richness, and Simpson’s evenness and diversity (recorded as 1 – D; Magurran 2004) for each taxonomic community at each sampling point for each visit. I summarized all community metrics per sampling point and later pooled metric data per trail or park for use in a series of 1 – way ANOVAs park (PROC GLM, α = 0.05 for all statistical analyses; SAS Institute Cary, North Carolina) with Tukey’s HSD (TUKEY in the means statement) to investigate differences in each taxonomic community among parks and between trails within each park. Data were transformed when found to violate the normality assumption and Kruskal Wallis tests with Dunn’s multiple comparison tests [R Version 3.1.2., http://www.r-project.org/;

dunn.test] were used when normality could not be attained. I also examined differences in species composition with the Bray-Curtis similarity index [R Version 3.1.2.,

http://www.r-project.org/; vegdist (dataset, method = “bray”)] among parks at the regional scale and between trails at the local scale (Su et al. 2004). The Bray-Curtis similarity index compares the shared species abundances between 2 areas and provides a

percent value of similarity ranging from 0 – 100%, with 0% representing complete dissimilarity and 100% representing complete similarity (Summerville and Crist 2003).

RESULTS